Recognizing 3D human motions using fuzzy quantile inference

摘要: Fuzzy Quantile Inference (FQI) is a novel method that builds simple and efficient connective between probabilistic fuzzy paradigms allows the classification of noisy, imprecise complex motions while using learning samples suboptimal size. A comparative study focusing on recognition multiple stances from 3d motion capture data conducted. Results show that, when put to test with dataset presenting challenges such as real biologically noisy" data, cross-gait differentials one individual another, relatively high dimensionality (the skeletal representation has 57 degrees freedom), FQI outperforms sixteen other known time-invariant classifiers.